The research effort supported by this award is aimed at preventing two major problems in continuous thin-slab steel casting: internal and surface cracks and whales ? bulges of the solidifying steel shell. A model-based spray cooling control system will be created to track, under disturbances and process model uncertainty, temperature profiles designed to satisfy quality constraints. Control of moving solidification front position will be addressed by developing theory of hybrid state estimation and control for systems with moving boundaries, applicable to a number of processes. The research approach progresses from the development of a controller/state estimator system for an idealized configuration to the real caster which has constrained spray-nozzle actuation, unmodelled process dynamics, and unreliable pyrometer measurements. Deliverables include fundamental control-theoretic results, controller and state estimator synthesis and analysis tools, software/hardware demonstration and validation on an industrial caster at Nucor Steel, an offline educational process control and visualization tool, documentation of research results in publications, and engineering student education.
Once completed, the results of this research will offer widely applicable, cost-effective control systems capable of accomplishing sophisticated spray cooling control tasks in continuous thin slab steel casting. This research will have broad impact across the entire industry due to quality improvement, (through fewer defects), energy savings (through yield improvement and lower reheating costs), better safety (from improved steel quality, and fewer breakouts and whale-defects), and competitiveness (by enabling the casting of new steel products). Improving spray cooling control to achieve even a one percent reduction in yield loss would save about $100 million per year (based on the roughly 100 million tons of steel produced each year in the U.S. and $100 per ton net cost of scrapping). Development of novel observation and control tools - sensor fusion with unreliable sensors and control of systems with moving boundaries - will address problems encountered in many other manufacturing processes. The research and educational infrastructure will be significantly enhanced through student involvement, industry and international collaborations, and broad dissemination of new knowledge.
This project was focused on designing breakthroughs automation for systems modeled by nonlinear partial differential equations in the steel industry, including flexible structures and solidification. These are systems that are extremely difficult or impossible to control with current methods. Through work funded by this project, algorithms have been developed that give guaranteed good performance. These algorithms are already being implemented in production at a steel mill in Alabama, and have been presented to others through industry and academic conferences, trade group meetings, and academic journals. The first new algorithm is a technique for controlling the temperature and shell thickness in a solidifying material. Both aspects can be simultaneously controlled to match a desired reference, a result never before achieved. Without such control, severe quality and safety issues may occur during steel casting. The proof is based on analysis of a nonlinear partial differential equation. The second new algorithm removes distortions of known frequency from oscillating machinery. It was applied to the problem of preventing a flexible beam from resonating. This was used to increase the oscillation frequency of a continuous steel caster mold, decreasing the size of oscillation marks on the steel surface. As a result, the machine is safer to operate and the resulting steel is less prone to cracks. Together, these projects will lead to significant cost savings, increased quality, and improved safety for the US steel industry. This project has also had a broader impact outside of steel production. These projects have funded multiple graduate students through Masters and Doctorates in Mechanical Engineering and Mathematics. Also, industrial case studies and models from this project have been used as the backbone of a revised undergraduate curriculum in signal processing and control at the University of Illinois. This case study has allowed students to learn and investigate the applications of the theory being taught in the course, better preparing them for the engineering challenges they will face after graduation. This new curriculum has been compiled into a textbook that will soon be published for other schools to use.
